Understanding The Risks From A Mains Disturbance When The UPS Has Failed  If a UPS has failed, the critical load is at the mercy of a mains failure. This is true for an On-Line UPS that has failed and is in bypass mode, and for an Off-Line UPS (including Line-Interactive models) that would fail to switch to battery if a power disturbance occurred. 

Relying on an automatic transfer to bypass is one way an On-Line UPS provides redundancy to protect a critical load. In many instances, the mains power quality is good enough that the risk of losing power to the critical load, while the UPS is being repaired, is acceptably low. In other cases, the risk of relying on the mains source for redundancy is unacceptably high. Off-Line UPS types, including Standby and Line-Interactive models, by design, connect the mains to the critical load, and only use the inverter to power the load when there is a mains disturbance. Therefore, a failed Off-Line UPS is essentially the same as an On-Line UPS that is on bypass. 
  
Mains Power Quality 

In general, the quality of the mains is deteriorating year to year due to greater demands for power. This is true in places like Australia, Europe, Japan and North America where the mains power quality is considered to be the best in the world, as well as in developing countries. The power utility companies spend much effort trying to provide a quality product, but there are many reasons, some beyond their control, that can cause power problems at a specific location. Some of these reasons are: 

  • Construction in the office or building
  • Construction in the neighborhood
  • Heavy industry nearby
  • Old wiring in the building
  • Old power distribution network
  • Inadequate power generating capacity
  • Natural causes (Animals, Storms, Floods, Lightning etc.)
A Definition Of Mains Reliability 

Many methods could be used to determine a reliability figure for the mains supply. One way is to determine the time the mains power was not within specification and to use this information to determine the mains availability. Availability can be determined as :
 
Availability1 = 
(Ttotal - Tout of spec) x 100% 
Where T = Time
Ttotal
 
1    If there are 52 disturbances in one year and each disturbance averages 1 second, the Availability is equal to ((31,536,000 - 52)/31,536,000)* 100% = 99.9998351% which is very close to perfect. However, each of the 52 disturbances would cause a failure of critical equipment! (There are 31,536,000 seconds in a year.)
 
For the purpose of determining the risk of a power interruption to the critical load, if the UPS is not functioning, availability is not a good figure to use. A better method for the purpose of assessing the risk of a critical load loss, is to count the number of power disturbances during a period of time (e.g. 1 year) and to set the MTBF as the mean (average) time. 
 
Mains MTBF = 
8760 Hours 
[1 Year = 
8760 Hours]
Number of Power Disturbances in One Year
  
The Mains MTBF calculation gives a clearer indication of the risk than does the availability figure because it eliminates the duration of the disturbances as a factor. The Mains MTBF can be converted into a mains failure probability (P) using: 
 
Pmains failure  = 1 - e-(UPS MTTR/Mains MTBF)
 
Table 1. Mains MTBF, Failure Probability and Availability
Frequency Of Disturbances Mains MTBF  Probability Of A Mains  
Disturbance In 24 Hours		 Mains Availability
1 disturbance per year  8760 hours  0.27%  99.9999968%
4 disturbances per year  2190 hours  1.09%  99.9999873%
12 disturbances per year  730 hours  3.23% 99.9999619%
52 disturbances per year  168 hours 13.28%  99.9998351%
365 disturbances per year 24 hours  63.21 %  99.9988425%
  
Table 1 shows the mains MTBF for several disturbance frequencies and a comparison with availability (assuming each disturbance is 1 second long and an MTTR of 24 hours). A reasonable approximation is that mains with good quality will have 4 to 10 disturbances a year. Medium quality would be 11 to 52 disturbances a year and poor quality would be when more than 50 disturbances happen each year. 

How Likely Is The UPS To Fail? 

Many factors contribute to the probability that a UPS will fail. The best indicator is MTBF, however, determining an accurate MTBF figure for a specific UPS is difficult. Modem UPS designs can demonstrate MTBF values in the range of 25,000 to 100,000 hours. The standard formula used to determine the probability of a failure give an MTBF value is: 

 
Pf  = 1 - e-(t/MTBF)

Table 2 shows the probability of failure in one year, and in three years, for several 
MTBF figures. 
 

Table 2. Probability of Failures
MTBF Pf in One Year  Pf in Three Years
8,760 hours (1 year)  63% 95%
25,000 hours (2.9 years) 30% 65%
50,000 hours (5.7 years) 16% 41%
100,000 hours (11.4 years) 8.4% 23%
200,000 hours (22.8 years) 4.3% 12%
400,000 hours (45.7 years) 2.2% 6.4%
 
Why UPS MTTR Is Important 
 
One way to quantify reliability is to estimate the likelihood that a UPS will be able to do its job and this is called Mission Reliability. However, the risk to the critical load cannot be truly assessed unless the mean time to repair (MTTR) of the UPS is considered. The longer the time to repair the UPS. the more risk there is that a mains failure will cause a critical load failure. Graph 1 shows the probability of a critical load failure versus mains MTBF for several MTTR values. 
 
Graph 1
 
It is interesting to note that if the UPS MTTR is 1 hour, the mains MTBF does not have a great effect on the probability of a critical load failure. However, probability of a critical bus failure can be significant if the MTTR is 24 hours or more and the mains MTBF is low. This matches our intuition. If it takes 48 hours to get the UPS running again, and there is a power disturbance almost every day, we would expect a power 
loss to the critical load. 
 
Ways To Reduce The Risk 

There are two main ways to reduce the risk associated with having your critical load powered by the mains when the UPS is being repaired 

    1. Redundancy 
    2. Low mean time to repair (MTTR)
Redundancy As A Method Of Mitigating Risk 

Redundancy generally means having a second UPS in reserve so if one fails the other one will continue to support the critical load. As long as one of the UPS units is working properly, when a power disturbance occurs, the risk of a critical load loss is minimized. 

A more economical means of achieving redundancy is to have a modular UPS with a 
redundant module. For example, if the critical load needs 5kVA, a modular UPS made up of six, 1kVA modules could be used. If one module fails, the five remaining modules are sufficient to power the load. 

However, if a single 5kVA UPS were to fail, the mains would support the critical load 
until the UPS is repaired. 

The "modular redundancy" solution is a more economical method to achieve redundancy because the cost of an additional module is much less than the cost of another UPS. UPS equipment that employs modular redundancy is now practical and cost effective in the 5 to 50 kVA range for true On-Line UPS equipment and in the 1 to 5kVA range for Line-Interactive equipment. 

Low MTTR As A Second Method Of Mitigating Risk 

A low MTTR means minimizing the UPS down time and can be achieved: 

  • If the power modules can be "hot-swapped",
  • And the modules can be changed by a "non-technical" person,
  • And a spare module is on site
Many UPS products are now "modular." However, if it is necessary to wait for a replacement module, the critical load is at risk while waiting for the module to arrive. The problem could be compounded if the module has to be replaced by a "qualified service technician" who may not be immediately available to do the repair. 

Fortunately, both true On-Line and Line-Interactive UPS are now available that offer an MTTR by a non technical person of 10 minutes or less (assuming a spare module is at hand). 

Conclusions 

To depend on the bypass in the event of a UPS failure, or during routine maintenance, exposes the critical load to a mains disturbance. Reducing the time on bypass reduces the risk of a mains disturbance causing a load failure. Never exposing a critical load to a mains disturbance is best, no matter how reliable the local mains supply is. 

To avoid using bypass, the best (and most cost effective) solution is to use a modular redundant UPS because, if a module fails, the UPS will continue to protect the critical load. The critical load is safe while the replacement module is being changed, even if the module has to be returned for repair. Modular redundant UPS equipment have the advantages of very low MTTR and of allowing a module to be safely changed by anyone.  For all critical applications, including file servers and UNIX workstations, modular redundancy is now the preferred solution.
 
Back to Technical Bulletin